What exactly are you modelling?

traveler_hauptman

I often see a big disconnect between what models actually are and how people try to use them. This is true for all types of models, but here it makes sense focus in on solid models for mechanical design capture. I'm really curious about what other peoples experience has been. 

A fundamental premise is that a model is an approximate representation of a real system that we use to make predictions or communicate about that system.

My assertion is that adding details to a model that don't support the goals of the model is a waste of time. This takes two forms. Details that don't support the end result (modeling internal details of an off-the-shelf part, bearing balls or grade markings on a bolt for instance). Details that are beyond the fidelity of the model such that changing the parameters of the detail has no effect on the end result. The latter is seen in simulation more often that solid modeling I think.

A discussion on how to create a knurl in Onshape got me thinking on this topic again. The OP wanted help creating a knurl and the reply created geometry that gave the appearance of a knurl. It seems that the answer satisfied the OP and everyone was happy. Everyone except me.

Traditionally a knurl is a deformation process. A tool rolls on a plastic surface forcing material out of the valleys of the pattern and onto the peaks. Which means that the knurled surface of a cylinder has a larger diameter than the starting surface. 

So what would I model for a knurl? In the design phase we care about collisions and I would model the knurl as a larger diameter cylinder to capture the potential for interference with other parts, especially parts being assembled over the knurl. In a manufacturing drawing, the knurl will be a drawing note rather than dimensioned in the drawing and the lines of the knurl cylinder will give the right information in the drawing views. Finally, when user manual or marketing materials need it, I would add the visual representation of the knurl.

To be clear, are reasons, due to manufacturing process or model intent that the knurl model that sparked this could have been exactly the right model for the job. 

The knurl above is just one example of differences between what a model is useful for and what we do with them. 

I think this is important because as soon as the Onshape guys loose sight of the fundamentals of modeling, we start getting a bunch of features that don't serve us, at the expense of the features we need. 

What do you think? What are the worst offences of useless detail you have seen? Can Onshape do something to help? 

nav

Hi @traveler_hauptman I'm one of the knurl guys, I totally agree with you some features and requests from the OS community do not need to be modeled, as you mention some of them are just a note on a mechanical drawing, however I have to admit that it was fun learning to model the knurl in OS as well as other weird non-mechanical challenges the community posts. I try to see odd requests as an opportunity to learn something new. I'm a Mechanical Engineer and I totally understand your post. OS being cloud based, free for the moment and so powerful, you can't help see all types of requests in the forums. Its the first time all people have access to such a powerful CAD tool online.

traveler_hauptman

@nav I agree. I tried to make it clear that the post was just inspiration for my thinking and that there were valid reasons for the post being as it was.

billy2

traveler-thanks for posting this.

When I read about the knurl I said to myself 'not on my projects'. I'd blow a gasket if someone added a knurled part to one of my projects.

I get it, a pretty bolt, gold plate it, print it out and hang it on the wall. 

I use 3D CAD (engineering layouts) to claim space or volumes. I use the minimal features to represent objects. 

When building a prototype, I have the layout open checking the prototype against the layout. I've had .01mm interference and sure enough the layout shows the same interference. I ask all the engineers to keep the layout true because it's everything.

If I get on to a project and there's no layout, I create one. It's the fundamentals to design.

We hire a lot of college engineers who know how to make an parts, assembly and drawings; but none of them understand the concepts of a layout to control a design. Pro/e & SW missed to boat, people struggle to control a design in these systems, they're not conceptual tools. Yet that's the best we have so far. Both these system work if you have a design and want to make 3D CAD geometry. 

Many designs start with a napkin sketch. Then you create a top down assembly? Try and get a college grad to understand all that. 

From a napkin sketch, to an engineering layout, to manufacturing control. How would you rate current CAD systems at achieving these steps?

3dcad

I use rendering software to represent the final details such as knurls, surface structures and others. It is not time well spent if I try to replicate wood structures in solid's surface